Image Restoration Model Using Dilation Gaps for Compact Camera Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image capturing technologies face challenges in maintaining image resolution and sensitivity while reducing the volume of capturing apparatuses, particularly in low illuminance environments, due to the trade-off between sensor size and focal length, leading to lower image quality in compact devices like smartphones compared to DSLR cameras.
Innovation Solution
The proposed method involves an image restoration technique using a convolutional neural network with dilation gaps, which rearranges compound eye vision images to enhance resolution by combining low-resolution images from multiple lenses, allowing for a thinner camera design while maintaining high image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If the size of the sensor decreases to reduce the volume of the capturing apparatus, then the volume is reduced, but the amount of light incident on the sensor decreases leading to lower image resolution and difficulty in low illuminance environments
Solution Approach 1:
The patent divides the capturing apparatus into multiple lens units (first lens unit, second lens unit, etc.) with different focal lengths and optical characteristics. Each lens unit captures images of the same object from different optical perspectives, allowing the system to combine these segmented captures to achieve high resolution without requiring a single large sensor
Solution Approach 2:
The patent combines multiple low-resolution images captured by different lens units to generate a single high-resolution image. The image processing unit integrates the optical information from multiple lenses, merging the data to compensate for the limited light-gathering capability of individual small sensors while achieving superior resolution
2Volume of moving object
If the size of the lens decreases to reduce the volume of the capturing apparatus, then the volume is reduced, but the focal length decreases leading to lower image quality
Solution Approach 1:
The patent segments the optical system into multiple lens units with different focal lengths (e.g., first lens unit with focal length f1, second lens unit with focal length f2 where f1 ≠ f2). This segmentation allows each small lens to have optimized focal characteristics while the combination provides equivalent or superior image quality to a single large lens
Solution Approach 2:
Each lens unit is designed with specific local optical properties tailored to its function. The first lens unit may be optimized for wide-angle capture while the second lens unit is optimized for telephoto capture, allowing each component to have high manufacturing precision for its specific purpose while the overall system maintains compact volume
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is an image restoration apparatus and method. The image restoration apparatus may store an image restoration model including a convolutional layer corresponding to kernels having various dilation gaps, and may restore an output image from a target image obtained by rearranging a compound eye vision (CEV) image by the image restoration model.